Gesture Recognition Input Method for AR

Nowadays, most of the input methods of the smart glasses are acoustic sensing, so we came up with the idea that we can use gesture recognition to new an input method. We use Leap Motion and Google Cardboard to simulate a smart glass with depth sensor, and SQLite as the database. Also, a windows computer used to receive the information of Leap Motion and interpret them into gestures. We now have six gestures which were trained by SVM to operate our device. The average accuracy has reached 80%(use cross-validation). Yet so far, we've only developed Traditional Chinese version, we're still looking forward to developing other versions with many different languages.

The gestures' using order has been described in the uploaded GestureOrder.pdf.